The promise of artificial intelligence in urban mobility has led to a surge in municipal procurement of advanced traffic management technologies. However, a significant phenomenon known as the deployment gap has emerged, reflecting the disconnect between the acquisition of smart city technologies and their actual operational deployment. Cities across North America and Western Europe have invested heavily in traffic AI systems, yet many of these technologies remain dormant or continue operating in rigid legacy modes. This discrepancy is not primarily the result of technological limitations but of systemic organizational, procurement, and operational barriers within municipal governments. This article examines the root causes of the deployment gap by analyzing why municipalities struggle to operationalize traffic AI despite already possessing the necessary technologies.
To systematically examine this issue, the analysis focuses on four major categories of traffic AI that are widely deployed but frequently underutilized: (1) Adaptive Traffic Signal Control Technologies (ASCT), (2) edge AI and computer vision cameras, (3) Vehicle-to-Infrastructure (V2I) communication modules, and (4) predictive analytics and digital twin platforms. Despite substantial public investment in these technologies, their operational deployment is often constrained by siloed organizational structures, legacy information technology environments, procurement inefficiencies, and persistent public-sector workforce shortages. Although these systems represent significant advances in intelligent transportation capabilities, their value cannot be fully realized without the institutional capacity required to integrate, maintain, and continuously optimize them.
The deployment gap represents a critical challenge for municipalities seeking to modernize transportation infrastructure and improve urban mobility outcomes. Beyond the financial implications of underutilized public investments, inactive or partially deployed traffic AI systems limit opportunities to improve roadway safety, reduce congestion, lower emissions, and strengthen transportation resilience. By examining the organizational, technical, and policy barriers that prevent municipalities from fully activating these technologies, this article provides a comprehensive assessment of the deployment gap and identifies practical considerations for moving from technology procurement toward sustained operational deployment.
The existing literature on intelligent transportation systems (ITS) and traffic AI deployment highlights a critical disconnect between technological capabilities and organizational readiness. A prominent theme across recent studies is the severe workforce challenge faced by state and local departments of transportation (DOTs). Research indicates that these agencies struggle significantly to implement artificial intelligence because most computer and data scientists prefer working in the private sector, making it exceedingly difficult for public transportation agencies to recruit and retain AI-knowledgeable labor1. This talent drain leaves municipalities ill-equipped to handle the complexities of modern traffic AI systems, forcing a reliance on legacy operational paradigms that fail to leverage new capabilities.
Empirical surveys further underscore the depth of this internal skills deficit. A recent comprehensive survey of transportation professionals revealed that 48.3% have received no formal artificial intelligence training2. Furthermore, 73.3% of respondents cited a lack of technical skills, and 66.7% identified a lack of hands-on experience as critical gaps preventing effective AI implementation, as summarized in Table 12. These statistics vividly illustrate why highly advanced, recently procured traffic management systems often default to their most basic, pre-programmed operational states. The advancement of ITS is expected to create new, highly specialized roles, such as transportation data scientists and AI/machine learning engineers, requiring specialized skills in big data analytics, machine learning, and geographic information systems3. However, the current municipal workforce, historically dominated by civil engineers, lacks the specialized skills in big data analytics and machine learning required to calibrate and maintain these advanced systems3. This disconnect highlights a fundamental flaw in the assumption that technology alone can drive smart city innovation. Without the human capital to interpret predictive analytics or fine-tune adaptive signal controls, the technology’s potential remains unrealized.
Table 1: Summary of AI Implementation Gaps in Transportation Professionals
| Gap Category | Reported Percentage | Primary Impact on Deployment |
|---|---|---|
| No Formal AI Training | 48.3% | Inability to calibrate advanced systems |
| Lack of Technical Skills | 73.3% | Over-reliance on external contractors |
| Lack of Hands-on Experience | 66.7% | Delayed operational activation |
In addition to workforce challenges, the literature extensively addresses the infrastructural and interoperability hurdles of deploying edge AI and V2I technologies. Smart city initiatives increasingly leverage edge computing to process distributed sensor data locally, which reduces network congestion and latency while enabling adaptive traffic control and rapid incident response4. However, edge artificial intelligence inference workloads require specialized infrastructure capable of supporting high power densities of 10 to 15 kW per rack, a specification that traditional municipal IT facilities and legacy traffic cabinets are typically inadequate to support5. Furthermore, integrating hardware and software from multiple vendors in these edge computing environments requires strict adherence to industry standards like MQTT and OPC UA to foster ecosystem interoperability and prevent vendor lock-in4. The literature suggests that without these foundational standards, cities risk building fragmented networks where individual AI cameras or sensors cannot communicate with the broader traffic management ecosystem, ultimately nullifying the benefits of the technology.
To examine the municipal deployment gap while maintaining methodological rigor, this study relies exclusively on secondary data sources, documented case studies, publicly available government reports, and municipal procurement records. Rather than generating new primary data through interviews or surveys, the research synthesizes existing evidence to identify recurring organizational, technical, and policy barriers that limit the operational deployment of traffic AI technologies. This secondary-data approach enables comparisons across multiple municipalities and transportation agencies while minimizing the biases associated with individual case observations.
A primary source of quantitative evidence is the U.S. Department of Transportation Intelligent Transportation Systems Deployment Tracking Survey, which provides national data on the adoption and operational status of intelligent transportation technologies across municipal jurisdictions.6 These findings are complemented by municipal technology assessments, workforce reports, and industry publications that document implementation challenges related to staffing, procurement, interoperability, and long-term operational support.3,6 Examining these sources collectively makes it possible to distinguish between technology acquisition and sustained operational deployment, thereby providing a broader understanding of the deployment gap.
To complement the broader literature, this study also considers the author’s previously published hybrid traffic-prediction framework (SSRN 5215361) as a representative example of deployable transportation AI. The framework integrates convolutional neural networks (CNNs), long short-term memory (LSTM) networks, and gradient boosting for traffic prediction using the METR-LA dataset, demonstrating more than a 30% reduction in prediction error compared with conventional approaches. Rather than evaluating the framework itself, the present study uses it as an illustrative example that technically mature traffic AI solutions already exist, allowing the analysis to focus on the organizational, procurement, infrastructure, and workforce barriers that continue to limit municipal deployment.
To strengthen the analysis, documented implementation case studies are used to illustrate how these challenges manifest in practice. One representative example is the City of Alexandria’s Phase III ITS Integration Project, which connected 58 existing traffic signal cabinets through a passive optical network architecture linking field infrastructure with the municipal traffic management center.7 Rather than treating individual projects as isolated success stories, these case studies are analyzed alongside broader national datasets to identify recurring patterns affecting deployment outcomes.
The collected evidence is interpreted through a qualitative thematic analysis organized around four recurring barriers identified across the literature: procurement and policy friction, internal workforce capability, vendor lock-in and interoperability, and departmental silos with funding misalignment. Structuring the findings around these recurring themes enables a systematic evaluation of why municipalities continue to struggle with operationalizing traffic AI despite substantial investments in intelligent transportation technologies.
The findings of this research explicitly structure the explanation for the deployment gap around four primary barriers. These barriers collectively explain why already-acquired traffic AI sits dormant in municipal networks, despite the clear potential for operational improvements.
Municipal procurement frameworks have traditionally been designed to acquire physical infrastructure rather than continuously evolving software platforms. As a result, AI applications often experience prolonged deployment delays even after the supporting hardware has been installed, creating a disconnect between technology acquisition and operational readiness. Recent federal policy initiatives seek to address these procurement challenges but also introduce additional compliance requirements. For example, in April 2025, the White House Office of Management and Budget (OMB) issued Memorandum M-25-22, establishing federal AI procurement requirements that mandate contract terms designed to prevent vendor lock-in and protect government data8.
In addition, the July 2025 AI Action Plan directed the General Services Administration (GSA) to develop an “AI procurement toolbox” to standardize federal AI acquisition and strengthen compliance with data governance requirements.9 Although these initiatives are intended to streamline AI procurement, municipalities continue to face overlapping state and local regulatory constraints. For example, Kansas enacted legislation in April 2025 prohibiting state agencies from procuring or using AI platforms controlled by foreign adversaries, joining at least 15 other states with similar restrictions.8 The cumulative effect of these layered procurement requirements is slower technology deployment, longer implementation timelines, and delayed operational activation of municipal traffic AI systems.
A second barrier is the shortage of specialized artificial intelligence expertise within municipal departments of transportation (DOTs). Historically, these agencies have been staffed primarily by civil engineers, creating a significant gap in data science, machine learning, and AI operations capabilities. State DOTs consistently identify the limited availability of machine learning expertise as a major obstacle to deployment, prompting recommendations to establish dedicated data science teams and use return on investment (ROI) metrics to strengthen leadership support for AI initiatives.10 Without these specialized capabilities, agencies often lack the expertise needed to interpret, calibrate, maintain, and continuously optimize AI-driven transportation systems.
Although educational institutions have begun responding to this skills gap, workforce development remains a long-term challenge. Recognizing that fewer than one in five job postings in the United States required a four-year degree as of early 2024, civil engineering educators are increasingly promoting stackable micro-credentials and incorporating AI-enabled technologies, including digital twins, into engineering curricula.11 However, these educational reforms will require time to influence the municipal workforce, leaving many agencies dependent on external consultants for AI implementation and maintenance. This dependence limits internal knowledge transfer, weakens long-term organizational capability, and reduces the sustainability of municipal AI initiatives.
Interoperability remains a major technical barrier because legacy traffic management infrastructure frequently relies on proprietary application programming interfaces (APIs) that limit integration with modern edge AI platforms. The diversity of proprietary APIs across cloud providers complicates data sharing, contributes to data silos, and reinforces vendor lock-in.12 As a result, municipalities often become dependent on closed technology ecosystems that make integrating new AI applications with existing traffic infrastructure both technically complex and financially costly.
To address these challenges, municipalities are increasingly adopting standardized middleware and open architectures. Embedded NTCIP translator cards provide standards-compliant communication between legacy field devices and modern traffic management systems.13 Likewise, the Advanced Transportation Controller (ATC) 5401 API Standard enables multiple software applications to securely share access to traffic controller interfaces and field devices, improving interoperability across heterogeneous transportation environments.14 The operational benefits of these approaches are substantial. In a Bucharest case study, an intelligent traffic control system using a hierarchical edge-cloud architecture and a YOLOv5 model reduced average vehicle waiting times by 37.5% while increasing traffic throughput by 38.9% during peak periods, as illustrated in Figure 1.15
Figure 1: Hierarchical Edge-Cloud Architecture for Traffic Control
Departmental fragmentation and funding misalignment represent the final major barriers to operational traffic AI deployment. Successful implementation requires sustained coordination among transportation, information technology, procurement, and finance departments, yet these functions often operate independently with differing priorities and budget cycles. As a result, many municipalities struggle to integrate AI systems into routine transportation operations despite having already invested in the underlying technologies. Effective digital transformation therefore depends not only on technology adoption but also on cross-functional collaboration, continuous information sharing, and organizational processes that support long-term system integration.16
Funding structures further reinforce this deployment gap. Capital investments frequently cover the acquisition and installation of AI technologies, while insufficient operational funding limits software maintenance, model updates, workforce training, and system optimization throughout the technology lifecycle. Although recent federal AI procurement initiatives encourage stronger governance and lifecycle planning, municipalities must also align long-term operational budgets with their technology investments. Without sustained organizational coordination and ongoing financial support, AI systems are likely to remain underutilized despite substantial public investment.
The deployment gap in municipal traffic management represents a critical bottleneck in the evolution of smart cities. Despite the widespread procurement of advanced technologies such as Adaptive Traffic Signal Control, edge AI cameras, and predictive digital twins, these systems remain largely dormant. This research has demonstrated that the root causes of this underutilization are not technical, but rather systemic and organizational. The primary barriers—procurement and policy friction, severe internal skills deficits, vendor lock-in, and entrenched departmental silos—create an environment where agile software solutions cannot thrive within rigid, hardware-centric municipal structures.
Addressing the deployment gap requires a fundamental shift in how municipalities approach technology acquisition and workforce development. Cities must reform procurement frameworks to accommodate continuous software integration and prioritize the recruitment of specialized data science personnel. Furthermore, insisting on open standards and interoperable middleware can dismantle the walled gardens created by legacy vendors. By systematically dismantling these four barriers, municipal departments of transportation can finally activate the traffic AI they already possess, realizing the promised benefits of safer, more efficient, and highly responsive urban mobility networks.Ultimately, the successful deployment of traffic AI is not merely a matter of purchasing the right software, but of cultivating the organizational readiness to utilize it effectively. Future research should focus on longitudinal studies tracking the success of municipalities that implement these targeted organizational reforms, comparing their operational outcomes against those that maintain traditional procurement methods.
By shifting the focus from procurement to operational integration, cities can close the deployment gap and transform their dormant technological investments into active, life-saving smart city infrastructure. The financial implications of this shift are equally profound. Billions of taxpayer dollars are currently tied up in underutilized technological assets across North America and Western Europe. Activating these systems will not only improve the return on investment but also restore public trust in municipal innovation initiatives. Ultimately, bridging the deployment gap is the essential next step in realizing the full potential of artificial intelligence in public transportation, moving beyond the hype of smart city concepts into the reality of data-driven urban management.
